The major purpose of this research is to develop new methods for the design and analysis of complex data that are encountered in cancer clinical trials. The research will focus on four topics. [unreadable] [unreadable] 1. Efficient methods will be developed to estimate the distribution of time to event with censored data for two-stage treatment policies where one treatment is given up front as induction therapy followed by maintenance treatment given only to patients after responding to the induction treatment. Tests comparing different treatment policies will also be developed. [unreadable] [unreadable] 2. Methods for estimating the effect that treatment duration has on response will be developed for situations where treatment duration may be terminated prematurely due to adverse events using observational data. [unreadable] [unreadable] 3. Efficient and robust methods will be developed for estimating the effect of covariates on death from a specific cause using a proportional hazards model when some of the cause of death information are missing. [unreadable] [unreadable] 4. We will demonstrate the inefficiency of the adaptive design for monitoring clinical trials and develop sequential methods that are uniformly better. [unreadable] [unreadable]